Tuesday 18 November 2014

Dublin Electricity Generation - An Analysis Part 1


The first four days of November was characterized by very high winds. This article will focus on the CCGT generators in the Dublin / East Coast region over this period and will be covered in two parts. Some of this will get a bit technical - but alas, that is the nature of the subject.

Figure 1 shows Forecast Wind and Actual Wind and the profiles of 4 units in the region:

Figure 1: Total Wind Generation and Forecast for the Republic (Eirgrid) and Profiles of Dublin Generators (SEMO)
1st Nov - 4th Nov, 2014. Two fossil fuel plant must be running at all times for voltage.

Dublin has the largest electricity demand in the country and as a consequence, there is a requirement for two generators to be running at all times in the region to maintain voltage control. So we can see that Dublin Bay is run continuously and either Huntstown or Poolbeg is run alongside. Great Island is been run in "reserve", ramped up to full output when required. For example, on the 3rd, when the projected wind output did not materialize, Great Island and Huntstown saved the day. On the 2nd, Huntstown came offline and was replaced by Poolbeg, fulfilling the two generator requirement. But Huntstown did not shutdown completely. We know this because a CCGT cannot come online instantly. It takes a few hours to start from "cold" as the steam builds up pressure. It would therefore not have been possible for it to step in on the 3rd, within an hour of the wind dropping off, unless it was running in reserve. The wind output dropped by two thirds (from 976MW to 335MW) within the space of 4 hours as demand rose on the Monday morning. There was not much margin for error.

So there was total reserve, that we know about, with a potential output of 870MW running behind this period of high wind (Great Island and Huntstown 2). The regulations state that there must be back-up generation, known as reserve, equal to the single largest in-feed in the system, which usually works out at 480-500MW. But this is the minimum amount of reserve required and is usually made up of different types of generators with different response capabilities. These units are then run offline on "part load" with a combined minimum load of 480-500MW, but they would be capable of ramping up to full load if required, like what happened here.

Figure 3 shows the Scheduled Output (known as Market Scheduled Quantity or MSQ) for Huntstown for the day in the blue line. It was due to run at 11am as demand hit about 3,500MW. So it would have been kept "warm" during the early morning, running in reserve. It is possible that one of the other generators could have suffered a "forced outage", which would also have led Huntstown to step in earlier than scheduled. But in this case, the wind dropped out as the weather system moved on. In the former case, about 400-450MW would have dropped out instantly. In the latter case, the wind drops out at a slower rate but with a potential larger drop in output. In this case, it dropped out over 4 hours, but with a total loss in output over the period equivalent to about 1.5 times that of a conventional power station. So when comparing the two events, there are pluses and minuses, but with both you need "back-up" to plug the gap, as can be seen in the red line. Huntstown gradually ramped up over the 4 hours eventually reaching near max output. It also ran for a couple of hours more than scheduled later in the day as wind output remained below forecast.

Figure 3: Huntstown 2 Scheduled Output Vs Actual Output (SEMO) - 3rd November 2014


Figure 4 shows the constraint payments in Euros that the unit received for running in the system earlier than scheduled. They received a large payment at 7am for "constraining on" and had to pay back a large payment at 11am for "constraining off" as it's output fell, with smaller payments received over the rest of the day. The overall net effect was additional income of € 63,000 on top of Energy Payments of € 240,000 for the day. The Energy Payments are based on the Scheduled Output and the constraint payments are to compensate for running the additional hours outside the schedule. The unit would also receive Capacity Payments, for making capacity available during the day, in effect compensation for running in reserve behind the wind to cover the staff and fuel costs involved (exact amounts unavailable at time of writing).


Figure 4: Huntstown Constraint Payments for 3rd Nov, 2014. Euros (€) on Vertical Axis with Hour on Horizontal Axis


There are no constraint payments or Scheduled output (MSQ) for Great Island as presumably it is still in testing mode.

The need for additional reserve ?


Forced outages work out at about one per month on average. It is very rare that two units will drop out on the same day. This occurred just once this year - on 13th September, when one unit each from Moneypoint and Tarbert failed to come online. So the TSO (Eirgrid) will have to make sure there is surplus dispatchable plant available when these unforeseen events occur. But with high wind penetration, such as the above, additional risk is built into the system. Forecast wind that does not materialize, leads to a very similar situation as a forced outage of a conventional plant - there is a loss in projected output that must be met by other forms of generation. These over-estimated forecasts occur on average about three times a month (based on Eirgrid All Island Wind and Fuel Mix Reports), with varying degrees of lost output as one might expect. Figure 5 shows the All Island Wind Generation (dark blue line) and Forecast (torquoise line) for May 2014. There are a couple of quite large variances between forecast and actual wind output. (Reports for October and November will be interesting to see when available)


Figure 5: All Island Wind Generation and Forecast May 2014
The risk of this type of forecast error and a forced outage occurring on the same day is greater than that of two forced outages occurring on the same day. This event occurred three times this year - on 24th February (Moneypoint and a 200MW drop in wind), 7th (Lanesboro and a 300MW drop in wind) and 27th April (Moneypoint and a 100MW drop in wind). Of course, one could argue that the wind producing zero or near zero output like it did for long periods in July and September is also the equivalent of a forced outage.

So there is a need to maintain surplus conventional (dispatchable) generation, in a system without wind. But during periods of high levels of wind penetration in the system, such as we had during October and November, Eirgrid must maintain more levels of reserve (above the minimum required), i.e. fossil fuel plant on standby (kept "hot" and burning fuel), because there is always the risk that an error in wind forecasting, such as the above, will coincide with a failure or failures in fossil fuel plant. Eirgrid's Mission Statement is simple - "to keep the lights on", and that means covering for every eventuality.

Part Two will look at Poolbeg and Dublin Bay CCGT plants in more detail.

On a side note, in relation to the other type of forecast error, i.e. where wind generation is higher than the forecast (also occurs quite a few times every month), this does not necessarily lead to any benefit in the system either. This is because the additional generation occurs outside the Scheduled Output for Wind. More on this in a future blog post.


No comments:

Post a Comment